Exercises: NDVI Analysis with Sentinel-2 Imagery

Last updated on 2026-10-20 | Edit this page

Overview

Questions

  • What is NDVI and what does it measure?
  • How do I calculate NDVI from Sentinel-2 bands in QGIS?
  • How do I interpret and visualize NDVI results?

Objectives

  • Understand what NDVI measures and why it is useful
  • Load individual Sentinel-2 spectral bands into QGIS
  • Use the Raster Calculator to compute NDVI from red and near-infrared bands
  • Apply pseudocolor symbology to interpret vegetation health
  • Compare NDVI results against satellite basemap imagery

Introduction

A Normalized Difference Vegetation Index (NDVI) is a widely used landscape metric that quantifies the health and density of vegetation using satellite sensor data. It works by comparing the reflectance of red light (which vegetation absorbs) and near-infrared light (which healthy vegetation strongly reflects). The formula is:

NDVI = (NIR − Red) / (NIR + Red)

NDVI values range from −1 to +1. Values near +1 indicate dense, healthy vegetation; values near 0 indicate bare soil or sparse cover; and negative values typically indicate water, clouds, or snow.

Global NDVI composite showing vegetation density across the Earth’s surface.
Global NDVI composite showing vegetation density across the Earth’s surface.

In this exercise we will use a Sentinel-2 multispectral image to calculate NDVI for a region in south-central Kansas.

Callout

ESRI Sentinel 2 Atlas

Take a moment to explore Sentinel-2 Land Cover Living Atlas. Take a note of its features!

Esri’s Sentinel-2 platform in the ArcGIS Living Atlas is a free, web-based tool that provides dynamic access to global, multi-spectral satellite imagery. It features a rolling 14-month archive of the best, most cloud-free scenes, which update daily at a 10-meter spatial resolution.

Step 1: Set Up Your Project

  1. Create a folder on your desktop called NDVI_Analysis (if you have not already done so from the setup page).
  2. Open QGIS, close any pop-ups, and go to Project → Save As. Save the project as NDVI_Project inside your NDVI_Analysis folder.

Step 2: Download and Extract the Sentinel-2 Image

  1. Navigate to the workshop’s shared resources Google Drive. Under Day 1_Session 3a: Basic raster functions, download the ZIP file: S2B_MSIL1C_20260617T170849_N0512_R112_T14SPG_20260617T203803.SAFE.zip

  2. Save the ZIP file to your NDVI_Analysis folder and extract it:

    • Windows: Right-click → Extract All
    • Mac: Double-click the ZIP file
The NDVI_Analysis folder with the extracted Sentinel-2 data.
The NDVI_Analysis folder with the extracted Sentinel-2 data.

Step 3: Load the Red and NIR Bands

Sentinel-2 images contain 13 spectral bands stored as individual files. For NDVI we only need two:

  • Band 4 (Red): T14SPG_20260617T170849_B04.jp2
  • Band 8 (Near-Infrared): T14SPG_20260617T170849_B08.jp2

To find them:

  1. In the Browser Panel on the left side of QGIS, navigate to: Project Home → S2B_MSIL1C_…SAFE → GRANULE → L1C_T14SPG_… → IMG_DATA
  2. Drag B04 and B08 into the Layers Panel.
Bands 4 and 8 loaded into QGIS.
Bands 4 and 8 loaded into QGIS.

Toggle each layer’s visibility using the checkbox next to its name to see how the two bands differ: Band 4 captures visible red light, while Band 8 captures near-infrared reflectance that is invisible to the human eye but strongly reflected by healthy vegetation.

Step 4: Calculate NDVI with the Raster Calculator

  1. From the menu bar, select Raster → Raster Calculator.
  2. In the expression box, enter the NDVI formula. You can double-click the band names in the Raster Bands list to insert them:
( "T14SPG_20260617T170849_B08@1" - "T14SPG_20260617T170849_B04@1" ) / ( "T14SPG_20260617T170849_B08@1" + "T14SPG_20260617T170849_B04@1" )
The Raster Calculator with the NDVI expression entered.
The Raster Calculator with the NDVI expression entered.
  1. Click the three-dot button next to Output layer and save it as NDVI_Output in your NDVI_Analysis folder.
  2. Click OK to run the calculation.

Once the output loads, you can right-click Bands 4 and 8 in the Layers Panel and select Remove Layer: they are no longer needed.

Step 5: Apply Color Symbology

The raw NDVI output appears in grayscale. To make vegetation patterns visible, we need to apply a color ramp.

  1. Right-click the NDVI_Output layer → Properties → Symbology tab.
  2. Change the Render type to Singleband pseudocolor.
  3. Set Interpolation to Discrete.
  4. Set the Color ramp to Greens.
  5. Under the table, set the Mode to Equal Interval, reduce the Classes to 3 and click Classify.
  6. Double-click each number in the Value column and change the three values to 0.05, 0.33 and 0.66 (QGIS keeps them in order, lowest at the top).
  7. Click Apply.

The column header reads Value <=: each value is the upper limit of its class. So the three classes are NDVI up to 0.05, from 0.05 to 0.33, and from 0.33 to 0.66. Pixels above the last value are left transparent; this image has almost none, but for other images change the last value to 1 so nothing disappears.

The symbology popup with three NDVI classes configured.
The symbology popup with three NDVI classes configured.

Step 6: Refine the Color Scheme

The green-only ramp shows the classification, but we can make interpretation more intuitive by matching colors to vegetation health levels.

NDVI reference scale showing the relationship between NDVI values and vegetation health.
NDVI reference scale showing the relationship between NDVI values and vegetation health.

Using the reference scale above as a guide:

  1. Reopen the Symbology tab.
  2. Double-click each color swatch to change it:
    • First class (up to 0.05) → Red: water, clouds and shadows, which have very low or negative NDVI
    • Second class (0.05 to 0.33) → Yellow: bare soil, buildings and roads, or sparse and dry vegetation
    • Third class (0.33 to 0.66) → Green: growing vegetation such as crops, grass and trees
  3. Click Apply, then OK.
The NDVI result with red-yellow-green symbology applied.
The NDVI result with red-yellow-green symbology applied.
Callout

Why are the values lower than on the reference scale?

This image is a Level-1C product: top-of-atmosphere values, stored with a fixed offset of 1,000 added to every pixel (Sentinel-2 products processed since January 2022). Computing NDVI straight from these numbers, as we did, gives somewhat lower values than the surface-reflectance NDVI that reference scales describe, which is why the classes here stop at 0.66. The pattern (water, built-up land, vegetation) is still clear. For measurements you would publish, use the Level-2A (surface reflectance) product and subtract the offset first.

Challenge

Exercise 1: Interpret the NDVI Map

Look at your NDVI output and try to answer the following:

  1. Can you identify Wichita (the largest city in the region)? What NDVI values dominate urban areas?
  2. Can you locate any lakes or rivers? What values do they show?
  3. Where is the densest vegetation: is it farmland, forest, or something else?

Write down your initial observations. We will check them against satellite imagery in the next step.

Step 7: Add a Basemap for Comparison

To verify your interpretations, add a satellite basemap underneath the NDVI layer.

  1. From the menu bar, select Plugins → Manage and Install Plugins.
  2. Search for QuickMapServices and click Install Plugin (if not already installed from Session 1A).
  3. After installation, a QMS panel should appear on the right side of your screen. Search for imagery and add Esri Satellite (ArcGIS/World_Imagery) to your map.

Now toggle the NDVI layer on and off in the Layers Panel to compare your NDVI classification against the actual satellite imagery.

The NDVI layer overlaid on satellite basemap imagery for comparison.
The NDVI layer overlaid on satellite basemap imagery for comparison.
Challenge

Exercise 2: Verify Your Observations

Compare your NDVI results against the satellite basemap.

  1. Were your initial observations from Exercise 1 correct?
  2. Did anything surprise you: areas you expected to be vegetated that were not, or vice versa?
  3. Can you identify any agricultural fields? How do their NDVI values differ from surrounding natural vegetation?
Key Points
  • NDVI uses the ratio of near-infrared and red reflectance to quantify vegetation health and density.
  • Sentinel-2 Bands 4 (Red) and 8 (NIR) are the inputs for NDVI calculation in QGIS.
  • Refer to this Multispectral Band Combinations website here to learn what the function of each combination is!
  • The Raster Calculator applies the NDVI formula on a per-pixel basis across the entire image.
  • Thoughtful color symbology (red → yellow → green) makes NDVI results immediately interpretable.
  • Comparing NDVI output against a satellite basemap helps validate your interpretation of the results.